Automated Document Classification Digital Solution

Trade Agreement: CETA / WTO-AGP / CPTPP / NAFTA / CFTA / FTAs with Peru / Colombia / Panama / Korea
Tendering Procedures: Suppliers on permanent list or able to meet qualification requirements
Competitive Procurement Strategy: Best Overall  Proposal
Comprehensive Land Claim Agreement: No
Nature of Requirements: 

Amendment 005 is raised to respond to questions from suppliers.


Amendment 004 is raised to respond to questions from suppliers.


Amendment 003 is raised to respond to questions from suppliers.


Amendment 002 is issued to revise the RFP documents, respond to questions from industry and publish the list of selected suppliers.

The following pre-qualified suppliers have been selected to submit a bid.

AltaML Inc.
Irosoft Inc.
Horoma AI
Data Performers Company Inc.
Oproma Inc.
SaS Institute Canada Inc.
Element AI 
Diligen 
IMRSV Data Labs Inc.
PriceWaterHouseCoopers LLP



************************************************************************
Amendment 001 is issued to publish the RFP documents.
************************************************************************

Automated Document Classification Solution

Public Works and Government Services Canada (PWGSC), also known as, Public Services and Procurement Canada (PSPC), is issuing a Notice of Planned Procurement on behalf of the Geological Survey of Canada of Natural Resources Canada (NRCan) for an upcoming Request for Proposal (RFP) for an Artificial Intelligence (AI) requirement using the newly established AI Source list under EN578-180001/B.

The following NPP is directed only to the qualified suppliers of the AI source list.

Suppliers under Bands 1, 2 and 3 are invited to provide feedback on this NPP.


1.  REQUIREMENT OBJECTIVE

NRCan seeks to use AI techniques to identify GEOSCAN entries relating to a particular ore system and extract the spatial coordinates of the best matches. A set of entries known to be related to the targeted ore system will be provided by NRCan as the training data. 
All of the science publications that the Geological Survey of Canada (GSC) at Natural Resources Canada (NRCan) produces is included within the open and freely available GEOSCAN database. At present, GEOSCAN includes approximately 80,000 entries in XML format with publication dates ranging from 1847 to present. Each publication entry within the database includes metadata fields such as title, subjects (or keywords), abstract, plain language summary, and location information (i.e., latitude, longitude, National Topographic System grid). Because a large proportion of research programs are focused on Canada’s natural resources, the GEOSCAN publication database represents an important repository for information regarding Canadian base, precious, and critical metal deposits (i.e., “ore systems”). Periodic, review-type scientific articles represent the traditional approach for consolidating the knowledge gained from ore system research. As part of the current SOW, we seek to develop new tools for interrogating this publication database using the latest advances in artificial intelligence (AI).
2.  INTENT OF NPP
Under this NPP, NRCan is first seeking for Industry feedback and validation on drafts documents associated with the solicitation to confirm marketplace capabilities and interest. In consideration of industry’s insights and other operational imperatives, NRCan will then issue a request for proposals (RFP). 

Below you will find a list of questions in Annex D for industry to provide feedback. The topics are as followings:

A. Desired Business Objectives and Project Scope (Annex A to NPP);
B. Statement of Work (Annex A to NPP); 
C. Evaluation Criteria (Annex B to NPP);
D. Use Cases (Annex C to NPP); and
E. Questions to solicit feedback from industry (Annex D to NPP).

The information provided will aid in increasing NRCan’s understanding of the capability, viability and availability of a solution. Industry perspective on the approach, business, cost of development, timelines and draft requirement documentation, are of particular interest.

We are requesting suppliers to provide feedback to PSPC no later than end of day October 18th, 2019.

3. KEY ESTIMATED RFP TIMELINES

Publication of the NPP:    October 3, 2019
NPP Feedback submission deadline:  October 18, 2019
Publication of RFP Solicitation:   TBD
Bidders’ Webinar:    TBD
Bidder self-Identification deadline:  TBD
Bid Closure:     TBD  

4. PSPC AGILE RFP EVALUATION APPROACH

This RFP solicitation involves a bid evaluation and a three-stage procurement process. 

Mandatory Stage 1
The specific requirements of the SOW include: 
A. Identify text patterns related to one targeted ore system using publication metadata within GEOSCAN;

B. Build predictive model to rank all GEOSCAN entries according to their degree of fit with the text patterns identified through the metadata. The predictive model will be based on the NRCan selected training set. Model accuracy and precision must be in the range of 70-100% based on cross-validation with the NRCan selected training set and manual validation with a smaller subset of GEOSCAN entries by NRCan;

C. Extract the identifying information, spatial coordinates, and degree of fit (i.e., predictive model confidence score) as an electronic data table (e.g., comma separated value file) for all GEOSCAN entries. The spatial coordinates will be used by NRCan to define areas of higher mineral potential for the targeted ore system. 

Optional Stage 2
Build and deliver an open-source and automated digital solution that could be reconfigured and/or modified (e.g., by selecting a new training set) by NRCan to train and apply predictive models for other ore systems in the future. This automated solution is required to:
A. Must work with GEOSCAN metadata in its original format so that new entries into the publication database can be included in a continually updating predictive model;
B. Must allow predictive model developed during Mandatory Stage 1 to be trained on a new training set by NRCan;  
C. Must be able to be used and maintained by NRCan.
  
Optional Stage 3
Many GEOSCAN entries correspond to scientific reports and/or maps in PDF format. A third optional stage of the SOW would be to apply the predictive model(s) developed during Stage 1 to search through the original scientific documents rather than only using the metadata hosted in GEOSCAN. The spatial coordinates of the best matches could then be extracted using the associated GEOSCAN metadata. Where appropriate, a subset of publications in PDF format could be provided by NRCan for the optional Stage 3 of this SOW. The automated digital solution delivered as part of Optional Stage 2 is not required to search through PDF documents.
The selection of proposal(s) for contracting does not constitute a guarantee on the part of Canada that a contract amendment will be awarded for Stage II.

5. ENQUIRIES AND RESPONSE 

All enquiries and responses to the questions in Annex D must be submitted in writing to the Contracting Authority no later than October 18, 2019.


23240-200239/A
Lessard, Peter
Telephone No. - (613) 850-7602 

Delivery Date: Above-mentioned

The Crown retains the right to negotiate with suppliers on any procurement.

Documents may be submitted in either official language of Canada.